2.2.4. recsys.utils.cy_utils.cy_calc_new_item_mean

recsys.utils.cy_utils.cy_calc_new_item_mean()

Calculate new item mean vector \(\mathbf{m}_{vj}^{*}\). The formula is described in the equation (8) of obi-CTR paper.

Parameters

• user_mean (ndarray, shape = (n_latent_dims,)) – User mean vector \(\mathbf{m}_{ui}\).

• item_mean (ndarray, shape = (n_latent_dims,)) – Item mean vector \(\mathbf{m}_{vj}\).

• item_cov (ndarray, shape = (n_latent_dims,)) – Diagonals of item covariance matrix \(\Sigma_{vj}\).

• emp_topic_freq (ndarray, shape = (n_latent_dims,)) – Empirical topic frequencies \(ar{\mathbf{z}}_j\).

• rating (int or float) – Rating given by the user \(i\) to the item \(j\).

• hypa_rat (int or float) – Prior of “rating” normal distribution \(\sigma_{r}^2\).

• hypa_eps (int or float) – Prior of “item latent offset” normal distribution \(\sigma_{\epsilon}^2\).

Returns

new_item_mean – New item mean \(\mathbf{m}_{vj}^{*}\).

Return type

ndarray, shape = (n_latent_dims,)